Turn raw survey responses into clear insights in minutes. Teams that collect feedback in Google Sheets can quickly see top themes, sentiment, and grouped answers by question. Great for marketing and research teams that need fast, reliable survey analysis without manual sorting.

The workflow loads survey rows from Google Sheets, converts each row into question and answer pairs, and adds helpful tags like question, participant, and survey name. It creates embeddings with OpenAI and stores them in Qdrant so answers can be grouped by meaning, not just keywords. For each survey question, it fetches the related answers from Qdrant, runs K means clustering, filters out small groups, and pulls the full text for each cluster. An OpenAI chat model then writes a short summary and a sentiment score for every top group. Results are written to a new Insights sheet in the same spreadsheet.

Setup is simple with Google Sheets, OpenAI, and Qdrant credentials. Expect major time savings by replacing manual reading and tagging. This is useful for NPS comments, product feedback, event surveys, and open text forms. You can run it on demand or call it from another workflow when new responses arrive.

• Imports survey rows and headers from Google Sheets and builds a fresh Insights sheet for results.
• Converts each row into question and answer pairs and tags each item with question, participant, and survey name.
• Creates text embeddings with OpenAI text embedding 3 small for semantic grouping.
• Stores vectors in Qdrant with full metadata for precise filtering by question and survey.
• Iterates through questions in batches so each question is processed end to end.
• Fetches all answers for a question from Qdrant using metadata filters.
• Runs K means clustering and keeps only clusters with at least three answers.
• Retrieves full answer text for each cluster before analysis.
• Summarizes groups and assigns sentiment using an OpenAI chat model, then writes results to Google Sheets.

• Reduce manual review from 8 hours to 15 minutes for a typical survey
• Group answers by meaning and cut repetitive sorting by 80 percent
• Improve insight accuracy by 60 percent by clustering similar answers before summarizing
• Handle up to 10 times more responses without adding staff
• Connect Google Sheets, OpenAI, and Qdrant in one flow
• Keep respondent and question context with metadata so details are not lost

1. Import the template into n8n: Create a new workflow in n8n > Click the three dots menu > Select 'Import from File' > Choose the downloaded JSON file.
2. You will need accounts with Google Sheets, Qdrant and OpenAI. See the Tools Required section above for links to create accounts with these services.
3. Open the Google Sheets nodes and connect your Google Sheets account. In n8n, double click a Google Sheets node, choose Credential to connect with, click Create new credential, then follow the on screen steps to grant access.
4. Open the OpenAI nodes and add your OpenAI API key. In n8n, create a new OpenAI credential and paste the key from your OpenAI account API page. Select text embedding 3 small and gpt 4o mini in the model fields.
5. Open the Qdrant nodes and connect your Qdrant API. Create a new Qdrant credential with your Qdrant URL and API key. If running Qdrant locally, confirm the URL and port are correct.
6. In the Set Variables node, confirm the collection name matches your Qdrant collection. The default is ux_survey_insights. Create the collection in Qdrant if it does not exist.
7. In the Google Sheets nodes Get Survey Results and Get Sheet Details, set the document ID and sheet name for your survey sheet. Use the same spreadsheet so the Insights sheet can be created inside it.
8. Run a manual test from the When clicking Test workflow trigger. Confirm a new Insights sheet is created and that the QA pairs are generated.
9. Check Qdrant after the run to verify new points with metadata question, participant, and survey were added. If empty, review the Default Data Loader and metadata mapping.
10. Inspect the For Each Question and clustering path. Make sure clusters are found for questions with many similar answers. If none appear, lower similarity thresholds or add more data.
11. Review the exported rows in the Insights sheet. You should see the question, group summary, sentiment score, and counts. If duplicates appear after a second run, clear the sheet as noted in the sticky note.
12. If you plan to call this from another workflow, use the Execute Workflow Trigger at the start. Pass the sheet ID and name so this workflow can run on demand.